Metals & mining · Case study
Improving grinding-mill throughput with machine learning.
A mining company used machine learning to identify operating settings that improved throughput at a constrained grinding mill.
Client context
The business context.
A mining company needed to increase throughput at a grinding mill that constrained the wider production process. Capital availability limited its options, and earlier analytical approaches had not resolved the bottleneck.
Hundreds of operating variables could influence throughput. Data came from different systems and formats, making it difficult to distinguish the important process drivers from noise and identify better combinations of settings.
The work
How the engagement was delivered.
Analyze the operating data
Examined a broad set of process inputs, including rotational speed, discharge pressure, grinding-media load and material hardness. Used the available data to investigate the drivers of throughput.
Model the drivers and test alternative settings
Applied neural-network and random-forest models in machine-learning simulations. Isolated the variables affecting throughput and evaluated combinations of operating parameters that conventional analysis had not identified.
Apply the improved settings
Changed the operating parameters using the findings from the simulations. The intervention focused on the existing grinding process rather than increasing its equipment capacity.
What changed
A 2% increase in grinding-mill throughput.
The grinding process achieved a reported 2% increase in throughput after operating at the settings identified through the AI simulations. The intervention improved performance at an existing bottleneck within the company’s capital constraints.
Our services
Related areas of support.
Your next decision
